Skip to content

Google Gemini¶

Google Gemini offers multimodal capabilities with the industry's largest context window (up to 2M tokens).

Stable

Quick Facts¶

Property	Value
Context	Up to 2M tokens
Free Tier	Yes
Models	Gemini 2.0, 1.5 Pro, 1.5 Flash
Streaming	Supported
Functions	Supported
Vision	Supported
Embeddings	Supported

Setup¶

Get API Key¶

Go to aistudio.google.com
Create or sign in with Google account
Get API key from API Keys section

Configure¶

Environment VariablePythonYAML

export GOOGLE_API_KEY="..."

from sentimatrix.config import SentimatrixConfig, LLMConfig

config = SentimatrixConfig(
    llm=LLMConfig(
        provider="gemini",
        api_key="...",
        model="gemini-1.5-pro"
    )
)

llm:
  provider: gemini
  model: gemini-1.5-pro

Available Models¶

Model	Context	Input Cost	Output Cost	Best For
`gemini-2.0-flash-exp`	1M	Free*	Free*	Latest, experimental
`gemini-1.5-pro`	2M	$1.25/1M	$5.00/1M	Long context
`gemini-1.5-flash`	1M	$0.075/1M	$0.30/1M	Fast, cost-effective
`gemini-1.5-flash-8b`	1M	$0.0375/1M	$0.15/1M	Ultra low cost

*Free tier with rate limits

Usage Examples¶

Basic Usage¶

import asyncio
from sentimatrix import Sentimatrix
from sentimatrix.config import SentimatrixConfig, LLMConfig

config = SentimatrixConfig(
    llm=LLMConfig(
        provider="gemini",
        model="gemini-1.5-flash"
    )
)

async def main():
    async with Sentimatrix(config) as sm:
        summary = await sm.summarize_reviews(reviews)
        print(summary)

asyncio.run(main())

Analyze Many Reviews (Long Context)¶

# Gemini 1.5 Pro can handle 2M tokens - analyze thousands of reviews at once
config = SentimatrixConfig(
    llm=LLMConfig(
        provider="gemini",
        model="gemini-1.5-pro"
    )
)

async with Sentimatrix(config) as sm:
    # Analyze 1000+ reviews in a single call
    insights = await sm.generate_insights(large_review_set)

With Vision¶

async with Sentimatrix(config) as sm:
    result = await sm.analyze_image(
        image_path="product.jpg",
        prompt="What sentiment does this product image convey?"
    )

Configuration Options¶

LLMConfig(
    provider="gemini",
    model="gemini-1.5-pro",

    # Generation settings
    temperature=0.7,
    max_tokens=8192,
    top_p=0.95,
    top_k=40,

    # Reliability
    timeout=60,
    max_retries=3,
)

Best Practices¶

Use Long Context for Bulk Analysis
- Gemini 1.5 Pro handles 2M tokens
- Analyze entire review datasets in one call
Start with Flash for Development
- Free tier available
- Fast responses
Use Pro for Production
- Better quality
- Larger context